Subscribe by RSS

Graduate Seminar Series

Mon., Jan. 23, 2023 3:30 p.m.

Location: CK 185 and Live Stream

Speaker: Sichen Liu

Title: Simplified methods for modelling correlated or dependent parameters in health economic evaluations (548 kB)

Abstract:

In economic evaluation, model parameters are often not independent. Al- though there are existing methods to handle the not independent parameters, it is often technically challenging to implement these methods for economic modelers. Therefore, the applied economic evaluations often do not incorporate the correlation or depen- dence of parameters in the modelling. We propose a small extension of the routinely used methods with several examples. These simplified methods can be easily applied.

We provide analytical proofs and propose simplified methods to handle the not independent parameters in some typical modelling scenarios. We demonstrate how to quantify the unknown correlation coefficients of correlated variables based on published summary statistics. We illustrate the use of univariate normal distributions based on results from in the multiple linear regression model and the Cox proportional hazard model, instead of generating the multivariate normal distribution data, to capture the heterogeneity of the population. We introduce a conditional probability method to handle more than two state transitions within a single cycle of a Markov model.

Live Stream: